Exploring bias in a generalized additive model for spatial air pollution data.

The estimation procedure was iterative, with a Newton-Raphson update of the unknown variance parameter computed at each iteration. To test for concurvity, as defined in Equation 4, we fit the model p = h(x,y) to these data, with h modeled as a loess function. The fitted function h was highly significant, with a squared correlation coefficient of 0.57 between p and h(x,y).

also in 132,142 other papers